The Journal of Neurosciaoco, June 1998, T3(6): 2477-2482

Associative Long-Term Potentiation in Piriform Cortex Slices Requires GABA, Blockade

Evan D. Kanterl and Lewis B. Haberiy1s2 ‘Neuroscience Training Program and 2Department of Anatomy, University of Wisconsin, Madison, Wisconsin 53706

Previous studies have demonstrated that NMDA-dependent, ability of this system (Wilson and Bower, 1988; Ambros-In- long-term potentiation (LTP) can be induced in both afferent gersonet al., 1990; Ketchum and Haberly, 1991). and intrinsic association fiber systems in the piriform (pri- Long-term potentiation (LTP) is a form ofactivity-dependent mary olfactory) cortex. In this report we demonstrate that an synaptic plasticity that has been studied intensively in the hip- associative form of LTP can be induced by coactivation of pocampalcortex asa model for a cellular mechanismof learning these two systems, which terminate on adjacent apical den- and memory (Blissand Lynch, 1988; Madison et al., 1991). The dritic segments of pyramidal cells. Potentiating stimulus trains commonly studied NMDA-dependent LTP is a long-lastingand were delivered to either afferent or association fibers, and -specificenhancement of synaptic responsesinduced by weak shocks, which were nonpotentiating when delivered brief high-frequency stimulation of excitatory pathways. The alone, were delivered to the other pathway. Under control possiblerelationship of LTP to associative memory is perhaps recording conditions where homosynaptic (single pathway) best suggestedby the particular form of LTP known as asso- LTP is consistently evoked, coincident application of these ciative LTP. In the associative LTP paradigm, a weak (non- stimuli failed to induce LTP of the weak shock response. potentiating) stimulus to one set of synaptic inputs can produce However, after local blockade of the fast, GABA,-mediated potentiation if paired with a strong (potentiating) stimulus to a IPSP, associative LTP was consistently produced in both second,independent set of synaptic inputs (Levy and Steward, directions. induction was blocked by o-P-amino+phos- 1979; Barrionuevo and Brown, 1983). The strong stimulus pro- phonovaieric acid, indicating that it is dependent on acti- ducesthe critical level of postsynaptic depolarization necessary vation of NMDA receptors. It is speculated that afferent and for LTP induction, which the weak stimulus alone is unable to association fibers are segregated on different dendritic seg- generate. The weak stimulus doesnot need to be tetanic; it can ments of pyramidal cells in pirlform cortex to allow regulation be a single volley (Gustafssonand Wigstrom, 1986). of associative LTP by way of centrifugal inputs that modulate NMDA-dependent LTP in both afferent and associationfiber the activity of GABAergic . pathways of piriform cortex haspreviously beendescribed (Jung [Key words: olfactory cortex, plasticity, inhibition, long- et al., 1990; Kanter and Haberly, 1990). This report demon- term potentiation, GABA, NMDA] stratesassociative LTP between the two fiber pathways, which can be induced only after GABA, blockade. Piriform cortex has several features that make it a particularly A preliminary report of theseresults has been published pre- attractive system for modeling associative,content addressable viously in abstract form (Kanter and Haberly, 1991). memory (Haberly, 1985; Haberly and Bower, 1989). These in- clude a highly distributed input, highly interconnected principal Materials and Methods cells, distributed positive feedback, and a spatially diffuse en- Ratsweighing approximately 200 gm weredecapitated under ether an- esthesia.A blockof brain containingpiriform cortex wasremoved and semblecode. Pyramidal in piriform cortex receive two slices500 pm thick wcrc cut pcrpcndicular to the cortical surface with excitatory inputs on adjacent segmentsof their apical : a Vibratome (Lancer)in oxygen-saturatedmedium at 0°C.Slices were affcrcnt fibers from the lateral olfactory tract (LOT) synapse allowed to recover for 2 hr at room temperature before recording. The distally in layer Ia, and intrinsic associationfibers synapsemore mediumwas 124 mM NaCI,5.0 mM KCI, 2.4 mMCaCI,, I .3 mM MgSO,, proximally in layer Ib (Price, 1973). This anatomical conligu- 26 mM NaHCO,, 1.2 mM KH,PO,, and IO mM ~@ucose,equilibrated with 95%0,, 5% CO,. ration, with distributed input (layer Ia) and distributed positive Recordingswere taken from submergedslices continuously perfused feedback(layer lb) exciting adjacentdendritic segments,suggests with oxygenatedmedium at 30°Cin the chamberdescribed by Tseng the possibility of associative interactions basedon coincident and Haberly (I 988). Extracellular field potentials were recorded using inputs from the two fiber systems.Associative synaptic plasticity glassmicroelectrodes broken to tip diametersof l-3 pm and filledwith 2 M NaCI.Stimulus trains and testpulses were delivered through bipolar betweenthe two pathways could contribute to the discriminative tungstenmicroelectrodes placed under direct vision in the lateral ol- factory tract (LOT) for stimulationof afferentfibers and in layer Ib for stimulationof associationfibers. Received June 9, 1992; revised Dec. 8, 1992; accepted Dec. 10, 1992. Testpulses were IO0 +ec shocks given at intervals of IO set for LOT This work was supportedby Grant NS19865 from the NINDS (L.B.H.) and stimulationand 20-30 set for layer Ib stimulation.The slowerrate for NIH Training Grant GM07507 (E.D.K.). layer lb was requiredto maintain a stableresponse amplitude in this Correspondence should be addressed to Lewis B. Haberly, Department of Anat- fiber system(Kanter and Haherly, 1990).Shock strength for test pulses omy, University of Wisconsin, 1300 University Avenue, West Loading Dock, was adjusted to give a response amplitude of approximately I mV for Madison, WI 53706. layer la and 2 mV for layer Ib (typically 30-50%of maximum).Pulses Copyright 0 1993 Society for Neuroscience 0270-6474/93/132477-06$05.00/O elicitingnear-maximal responses were used for potentiatingtrains. Po- 2478 Kanter and Haberfy l Associative LTP in Piriform Cortex

STIM REC BIG STM Results A schematic diagram of the recording configuration is shown in Figure 1. Afferent fiber synapsesin layer Ia were activated by the stimulating electrode in the LOT. Association fiber syn- apseswere activated by the stimulating electrode in the middle to deep part of layer Ib. The recording electrode near the layer lb ‘7 Ia/Ib border registered the active part of the dipole for associ- + ation fibers and the passive part of the dipole for afferents, __-___I 1 .------t--- resulting in afferent and association fiber responsesof opposite II /\ P polarity. ,,, smTw------)-- Potentiating (strong) stimuli were delivered to tither afferents or association fibers and single (weak) pulseswere delivered to the other pathway. Two controls were necessaryin order to clearly establishthat the LTP induced wasassociative in nature. A potentiating train to the strong pathway alone should not Figure 1. Schematic diagram of recording configuration. P, superficial induce LTP in the weak pathway. Ten test pulsesto the weak ; plus signs indicate excitatory ; Roman numerals denote cortical layers; BK, bicuculline-containing pipette. pathway at 200 msec intervals should also have no effect. Under the same conditions in which homosynaptic (single pathway) LTP can be consistently evoked (Kanter and Haberly, tentiating trains consisted of 10 sets of four pulses at 100 Hz delivered 1990), associative LTP could not be induced by pairing strong at 200 msec intervals. The 200 msec interval approximates the rate of expl,oratory sniffing in the rat, which corresponds to the limbic theta and weak stimuli. Occasionally, a small, brief potentiation of rhythm (Macrides et al., 1982). This stimulus paradigm is highly effec- the association fiber responsewas seen(as in Fig. 3B), although tive for inducing LTP (Larson et al., 1986). During pairings of strong in most casesthis was not present. Paired stimulation in normal and weak stimuli, weak shocks (equivalent to test pulses) were delivered medium never produced any measurableeffect on the afferent at 200 msec intervals between the second and third pulses of the four- fiber response.Strong inhibitory mechanismsare present in pir- pulse bursts. Amplified responses were digitized and stored by com- iform cortex, as in other types of cortex (Scholfield, 1973; Satou puter; calculation of the slope of the rising phase was automated. In experiments where 2-5 PM bicuculline methiodide was bath ap- et al., 1982; Tsengand Haberly, 1988). SinceGABA antagonists plied, Ca*+ was elevated to 10 mM to block polysynaptic activity, and can facilitate LTP in the hippocampus (Wigstrom and Gustafs- KH,PO, was eliminated to prevent precipitation. In experiments using son, 1983, 1985;Gustafsson and Wigstrom, 1986),it was thought standard medium, 5 mM bicuculline was focally applied either from the that blockade of inhibition with bicuculline, which blocks the recording pipette or from a second pipette inserted near the recording fast, Cl--mediated IPSP in piriform cortex, might enable in- site. Bicuculline-containing pipettes were broken to tip diameters of IO- duction of associative LTP in this system. 20 pm, and when necessary, positive iontophoretic currents up to 20 Two methods were used to introduce bicuculline. In initial nA were used to pass bicuculline. n-2-Amino-5phosphonovaleric acid experiments (10 slices) bicuculline (2-5 PM) was bath applied. (D-APV), 3-amino-2-(4chlorophenyl)-propylphosphonic acid (phaclo- fen), and 3-amino-2-(4chlorophenyl)-2-hydroxy-propylsulfonic acid (2- In normal bathing medium, this resulted in epileptiform burst- hydroxysaclofen) were bath applied. Bicucuhine was from Sigma; D-APV ing activity, characterized by large all-or-none responseshaving was from Cambridge Research Biochemicals; phaclofen and a distinct threshold. In order to prevent epileptiform responses, 2-hydroxysaclofen were from Research Biochemicals, Inc. it was necessaryto block polysynaptic transmission, an accept-

160 1 ..*.. .:: -.:.. :: ..*..* c. . . ::. 5 130- E 0 v, . ..- ..*.*. .-:.: *.-.:- ET loo-‘.’ *....* .:- .: . . . . . El A’ f A A s+w W s+w

Figure 2. Associative LTP induction after bath application of bicuculline with elevated calcium. Potentiation of association fiber response by burst stimulation of afferent fibers is plotted as a function of time; amplitude is expressed as slope of rising phase of field EPSP; baseline response is normalized to 10096. Bar denotes period of bath application of bicuculline (BIG). S, strong potentiating trains to afferent fibers alone; W, weak shocks to association fibers alone; S+ IV, paired stimulation. Test pulses were delivered every 30 set; each point is an average of four consecutive responses. Extracellular calcium was elevated to 10 mM to prevent epileptiform activity induced by bath-applied bicuculline. Superimposed field potential responses are shown at right, 5 min before and 20 min after paired stimulation for control condition (before application of bicuculhne; top traces) and in the presence of bicuculline (bottom Iraces). Each trace is an average of six consecutive responses. Shock artifacts are truncated. Calibration: 2 msec, I mV. The Journal of Neuroscience. June 1993, U(6) 2478

A 140, BIG

BIG Figure 3. AssociativeLTP induction after focal aoolication of bicuculline in normal bat&g medium. A, Potentia- 1 tion of afferent response by burst stim- & ulation of association fibers: slope of rising phase of field EPSP versus time, baseline response normalized to 100%. B 160 Bar denotes period of focal application 1 of bicuculline (BIG) near the recording I.__1 site. S, strong potentiating trains alone; W, weak shocks alone; S+ W, paired Control stimulation. Test pulses were delivered every 10 set; each point is an average of eight consecutive responses. Aver- aged response traces (right) are ar- --I ranged as in Figure 2. B, Potentiation - of association fiber response by burst stimulation ofafferents. Test pulses were delivered every 30 set; each point is an I I I I I ' I BIG 60 100 120 140 160 average of four consecutive responses. k J Calibration: 2 msec, 1 mV.

able approach since associative LTP induction in this paradigm od, robust associative LTP was consistently observed without requiresonly monosynaptic activation of Ia and Ib fibers. This epileptiform activity. was done by usinga high concentration of CaZ+in the medium, In the examples shown (Fig. 3), bicuculline was introduced which elevatesthreshold without interfering with synaptic trans- through a second pipette near the recording site. Paired stim- mission (Miles et al., 1988). Increasing the Caz+ concentration ulation without bicuculline did not produce LTP, while the same to 10 mM suppressedpolysynaptic transmissionenough to pre- paired stimulation was effective at inducing LTP after insertion vent epileptiform activity effectively. of the bicuculline-containing pipette. Control stimuli to afferents Figure 2 illustrates an experiment of this type. Layer Ia af- alone and to association fibers alone produced no potentiation ferents were usedas the strong pathway and layer Ib intrinsic in the presenceof bicuculline, indicating that the LTP observed association fibers as the weak pathway. CaZ+ was elevated was associative in nature. throughout this recording. Neither trains to afferents alone nor Associative LTP was reliably produced in both directions: 5 Hz weak shocksto association fibers alone produced poten- afferent responseswere potentiated by strong stimulation of tiation of association fiber responses.LTP was produced only associationfibers paired with weak stimulation of afferents (Fig. when the stimuli were given simultaneously and only when 3A; 12 of 12 slices) and association fiber responseswere po- bicuculline was usedto block GABA,-mediated inhibition. The tentiated by strong stimulation of afferents paired with weak example in the figure shows a potentiation of approximately stimulation of association fibers (Fig. 3B; 15 of 16 slices).The 50% above baseline,which was the greatestamount of associa- magnitude of associative LTP in each pathway was similar to tive LTP observed with this technique. Induction of associative that of homosynaptically induced LTP (Kanter and Haberly, LTP was seenin 5 of 10 sliceswith this approach. 1990). A potentiation of 1O-l 5% was typically obtained for the Concernsabout activation of nonphysiological processesin afferent pathway from a single stimulus train, with saturation high Ca2+led to the development ofa secondmethod for block- after multiple trains at 15-25%. A potentiation of 2040% was ing GABA, inhibition without initiating epileptiform activity. typically obtained for the associationfiber pathway from a single Focal, iontophoretic application of bicuculline either from the train, with saturation at 50-75O/6.The time course was also recording pipette or from a second pipette inserted near the similar to that of homosynaptic LTP, with a more rapid onset recording site allowed the use of standard medium. Since GA- of potentiation seenin afferent fibers, and an early, decremental BA, blockade occurred only at the recording site and the re- phase seen in association fiber potentiation. Bicuculline pro- mainder of the slice was unaffected, epileptiform bursting was duced no apparent increasein the magnitude of homosynapti- avoided, perhapsby keepingthe population ofdisinhibited cells tally induced LTP in either pathway. below a critical number required for reverberating positive feed- Given the ability ofa GABA, antagonist to enable associative back (Traub and Wong, 1983). Current injection could be used LTP induction, we examined whether GABA, blockade has a to regulate the amount of bicuculline passing into the slice, similar effect on plasticity. In the presenceof the GABA, an- although in most experiments diffusion of bicuculline was suf- tagonistsphaclofen (two slices)or 2-hydroxysaclofen (two slices), ficient to block inhibition without any current. Using this meth- no associative LTP was observed. Associative LTP could sub- 2480 Kanter and Haberly l Associative LTP in Piriform Cortex

A 140, Discussion The present resultshave demonstratedthat associativeLTP can occur in both directions between afferent inputs on distal apical *..:..* ...... ::* . ..*.*. . . . . dendrites and association fiber inputs on proximal apical den- D-APV . . . ” drites of pyramidal cells in the piriform cortex. Like homosy- v, naptic LTP in the same fiber systems, induction of this asso- * *..-A. I ..*.*.-... ‘=‘..... El , 00 -:.‘.: .. . . .‘,...~~‘“..‘.‘. . -:.-.. : “.M.*.. . -*. .. . . ciative LTP was dependent on activation of NMDA receptors. El \ *. However, in contrast to homosynaptic LTP that can be elicited A A A AAA in standardbathing medium, associative LTP was only observed s w s+w s w s+w 80 I I II I 1 I 1 I 1 I when GABA, synaptic transmission was blocked during the 0 30 60 90 120 150 180 potentiating stimulus trains. Minutes Mechanism for modulation of associative LTP by inhibitory processes 0. . . The NMDA dependenceand similarity in magnitude and time . . *.. course suggestthat the LTP induced by temporally associated . . . ..:*... . activity in afferent and association fiber systemsis equivalent D-APV -...... *... to that evoked in either fiber system alone. An important ques- (II4 *.-:...... * *.. tion is, therefore, why blockade of GABA, receptorsis required fy+100 ...... *.... ***-*--**- ...... for associative but not homosynaptic induction. k From work on hippocampal cortex (Gustafssonand Wigs- A AAA tram, 1986; Kelso et al., 1986; Sastry et al., 1986; Wigstrijm et t- s+w S w s+w al., 1986; Gustafsson et al., 1987) it has been postulated that 80( , , , , I I I I I I I 0 20 40 60 80 100 120 140 associativeLTP of the form observed in the presentstudy occurs Minutes when depolarization from the strong stimulus train spreadsto Figure 4. D-APV blocks induction of associative LTP in both path- the location of synapsesactivated by the weak stimulus, thereby ways. A, Potentiation of afferent response by burst stimulation of as- allowing activation of NMDA receptorsthat are largely blocked sociation fibers: slope of rising phase of field EPSP versus time, baseline by MgZ+ at resting membrane potential. Consequently, it can response normalized to 100%. Bar denotes period of application of 15 be postulated that blockade of GABA, transmission increases PM D-APV. Bicuculline was introduced focally throughout the experi- ment by diffusion from the recording pipette. S, strong potentiating the level of depolarization attained in the weakly stimulated trains alone; W, weak shocks alone; S+ W, paired stimulation. Test synaptic zone in two possibleways: by facilitating the electro- pulses were delivered every 10 set; each point is an average of eight tonic spread of potential and/or by increasing the level of de- consecutive responses. B, Potentiation of association fiber response by polarization achieved in the strongly stimulated zone. The in- burst stimulation of afferents. Test pulses were delivered every 20 set; each point is an average of four consecutive responses. creasein electrotonic spreadof EPSPswould be a consequence of blocking the Cl- conductance, thereby increasing the mem- brane resistanceand length constant. sequently be induced in these slicesupon introduction of bi- The induction of an NMDA dependent slow component and cuculline. The concentrations of phaclofen (500 PM) and its buildup during potentiating trains (Fig. 5) is consistent with 2-hydroxysaclofen (250 FM) used were sufficient to block re- an increasedlevel of aepolarization in the zone receiving the versibly the intracellularly recorded slow, K+-mediated IPSP strong stimulus train. Blockade of the fast, Cl--mediated IPSP in piriform cortex (Hoffman and Haberly, 1989; E. D. Kanter would also be expected to increase the amplitude of the non- and L. B. Haberly, unpublished observations). NMDA-mediated EPSP component. Amplitudes of concomi- It has previously been demonstrated that the induction of tant EPSPs would be reduced by this IPSP in spite of its de- homosynaptic LTP in both afferent and association fiber path- polarizing polarity in piriform cortex (Scholfield, 1978), since ways is dependent on NMDA receptors (Kanter and Haberly, the Cl- equilibrium potential is at a hyperpolarized level with 1990). To determine if associative LTP in these pathways has respect to the peak of all but the smallest EPSPs. a similar dependence,the selective NMDA receptor antagonist Given that the conductance increasesassociated with GA- D-APV was applied in the bathing medium during stimulus BA,- and GABA,-mediated IPSPsin piriform cortex are rough- trains (Fig. 4). Induction in both directions was blocked by 15 ly comparable as measuredat the cell body, and that the GA- PM D-APV (three slicesfor each direction), indicating that the BA,-mediated IPSP is more strongly hyperpolarizing (Tseng associative form of potentiation is also dependent on NMDA and Haberly, 1988), it is surprisingthat associativeLTP in apical receptors. dendrites can be induced with GABA,, but not GABA, antag- Figure 5 illustrates the effect of bicuculline on responsesdur- onists. In addition, experimentsusing local application of GABA ing burst stimulation. When four-pulse, 100 Hz bursts were agonistsor blockers of synaptic transmissionsuch as Cd2+have recorded extracellularly in normal medium, the time course of shown that a large component of the GABA, responseis gen- the field EPSP wasrapid enoughso that there was no summation erated at the cell body, while the GABA, responseis generated of responses.After introduction of bicuculline a slowerresponse predominantly in the dendrites (Newberry and Nicoll, 1985; component appeared that summated over the burst. This slow Tseng and Haberly, 1988). component was blocked by 15 PM D-APV, suggestingthat it is Several factors could contribute to this finding. First, the ex- an NMDA receptor-dependentcomponent that is normally sup- periments on spatial localization of IPSPs do not rule out the pressedby GABA, inhibition. presenceof a significant GABA, component in the dendrites, The Jcumal of Neuroscience, June 1993, f3(6) 2401

r---Control BIC

BIC Control I BIG, APV BIC, APV Figure 5. Effect of bicuculline on field potentials during potentiating rtrains: responses to burst stimulation (four pulses at 100 Hz) of association fibers in layer Ib. Shock strength was adjusted to give response amplitudes of 4.0 mV for the initial peak. A, Response under control conditions. R, Focal application ofbicuculline (BIG) resulted in a long-lasting depolarization. C, Bicuculline effect was blocked by 15 PM D-APV. D, Superposition of A and B. E, Superposition of B and C. F, Superposition of A and C. Calibration: IO msec, 1 mV. since a dendritic component would have been reduced in am- ever, in view of the need for strong convergent excitation to plitude by electrotonic spread to the somatic recording site. exceed the threshold for LTP induction, it seemsreasonable to Second, activation of presynaptic GABA, receptors reduces postulate that disinhibition is required in vivo as well. GABA releasein hippocampus, leading to decreasesin both In the hippocampus, a variety of neurotransmitter systems GABA, and GABA, responses, and by this mechanism facih- have been implicated in the modulation of GABAergic inhi- tates LTP (Davies et al., 1991; Mott and Lewis, 1991). If pre- bition, including norepinephrine(Andreasen and Lambert, 199 1; synaptic GABA, receptors are also present in pitiform cortex, Doze et al., 199l), 5-HT (Ropert and Guy, 1991), enkephalin then GABA, antagonists could increase GABA,-mediated in- (Madison and Nicoll, 1988; Lupica and Dunwiddie, 199 l), so- hibition in addition to blocking the slow IPSP. The increased matostatin (Scharfman and Schwartzkroin, 1989), and ACh (Pi- GABA,-mediated inhibition could have therefore compensated tier and Alger, 1992). Anatomically, physiologically, and phar- for the decreasedpostsynaptic GABA, inhibition. Finally, the macologically distinct subpopulations of GABAergic neurons slow time course of the GABA,-mediated IPSP (onset at 30- have been describedin piriform cortex (Haberly, 1990; Sheldon 50 msec and peak at 150-250 msec under present conditions) and Aghajanian, 1990) that could serve as targets for this mod- would preclude a substantial modulatory role during the first ulatory action. burst-weak shockpairing, and fatigue that occursat a repetition An intriguing possibility is that the segregationofafferent and rate of 5 Hz could have limited its strength during subsequent association fiber inputs to different dendritic regions servesto pairings. allow associative plasticity to be modulated by centrifugal path- ways that terminate on GABAergic intemeurons. In the absence Functional significance of such segregation,associative LTP would presumably occur On the behavioral level, memory storageis modulated according without disinhibition. Consistent with this idea is the demon- to the “significance” of incoming sensory information. In the stration that, in the temporodentate pathway from entorhinal LTP model, such regulation of plasticity is necessaryto avoid cortex to dentate , associativeinteractions between crossed saturation of synapses.Control of inhibition is one possible and uncrossedprojections arc depcndcnt on the degreeof spatial mechanismwhereby modulatory neural inputs could determine overlap of synapses(White et al., 1990). whether a synaptic network is in a learning-receptive state. GA- Piriform cortex and hippocampalcortex sharesimple, orderly BA, transmissioncould serve as a “final common pathway” on structural fcaturcs that qualify them as model cortical systems. which several systems might converge in the regulation of syn- An advantage of piriform cortex over hippocampus is that as a aptic plasticity. primary sensory cortex, it lends itself well to investigation of It is conceivable that levels of inhibition evoked by natural the role of synaptic plasticity in memory storage and discrim- odor stimuli are less than those elicited by the synchronous inative processes(Haberly and Bower, 1989). The piriform cor- activation of fibers in vitro, and therefore that disinhibition is tcx may therefore be a useful system in which to investigate not required for the induction of associative LTP in vivo. How- how synaptic plasticity relatesto circuitry and system function. 2482 Kanter and HatMy l Associative LTP in Piriform Cortex

References Madison DV, Malenka RC, Nicoll RA (199 I) Mechanisms underlying long-term potentiation of synaptic_ transmission. Annu Rev Neurosci Ambros-lngerson J, Granger R, Lynch G (1990) Simulation of paleo- 14:579-397. cortex performs hierarchical clustering. Science 247: 1344-l 348. Miles R, Traub RD, Wong RKS (1988) Spread of synchronous firing Andreasen M, Lambert JDC (1991) Noradrenaline receptors partic- in longitudinal slices from the CA3 region of the hiowcampus.-- - J ipate in the regulation of GABAergic inhibition in area CA1 of the Neurophysiol 60:1481-1496. rat hiooocamnus. J Phvsiol (Land) 439:649-669. Mott DD, Lewis DV (199 1) Facilitation of the induction of long-term Banionuevo G,‘Brown TH (1983) Associative long-term potentiation potentiation by GABA, receptors. Science 252: 17 18-l 720. in hippocampal slices. Proc Natl Acad Sci USA 80:7347-7351. Newberry NR, Nicoll RA (1985) Comparison of the action of baclofen Bliss TVP, Lynch MA (1988) Long-term potentiation of synaptic with -r-aminobutyric acid on rat hippocampal pyramidal cells in vitro. transmission in the hippocampus: properties and mechanisms. In: J Physiol (Lond) 360:161-185. Long-term potentiation: from biophysics to behavior (Landfield PW, Pitler TA, Alger BE (1992) Choline@ excitation of GABAergic in- Deadwyler SA, eds), pp 3-72. New York: Liss. temeurons in the rat hippocampal slice. J Physiol (Lond) 450: 127- Davies CH, Starkey SJ, Pozza MF, Collingridge GL (1991) GABA, 142. autoreceptors regulate the induction of LTP. Nature 349:609-6 11. Price JL ( 1973) An autoradiographic study of complementary laminar Doze VA, Cohen GA, Madison DV (199 1) Synaptic localization of patterns of termination of afferent fibers to the olfactory cortex. J adrenergic disinhibition in the rat hippocampus. 6:889-900. Comp Neural 150:87-108. Gustafsson B, Wigstrom H (1986) Hippocampal long-lasting poten- Ropert N, Guy N (199 1) Serotonin facilitates GABAergic transmission tiation produced by pairing single volleys and briefconditioning tetani in the CA 1 region of rat hippocampus in vitro. J Physiol (Lond) 44 1: evoked in seoarate afferents. J Neurosci 6: 1575-l 582. 121-136. Gustafsson B, Wigstrom H, Abraham WC, Huang Y-Y (1987) Long- Sastry BR, Goh JW, Auyeung A (1986) Associative induction of post- term potentiation in the hippocampus usingdepolarizingcurrent puls- tetanic and long-term potentiation in CA1 neurons of rat hippocam- es as the conditioning stimulus to single volley synaptic potentials. J pus. Science 232:988-990. Neurosci 7:774-180. Satou M, Mori K, Tazawa, Y, Takagi SF (1982) Two types of post- Haberly LB (1985) Neuronal circuitry in olfactory cortex: anatomy synaptic inhibition in pyriform cortex of the rabbit: fast and slow and functional imnlications. Chem Senses IO:2 19-238. inhibitory postsynaptic potentials. J Neurophysiol48: 1142-I 156. Haberly LB (1990) =Olfactory cortex. In: The synaptic organization of Scharfman HE, Schwartzkroin PA (1989) Selective depression of the,brain (Shepherd GM, ed), pp 317-345. New York: Oxford. GABA-mediated IPSPs by somatostatin in area CA1 of rabbit hip- Haberly LB, Bower JM (1989) Olfactory cortex: model circuit for study pocampal slices. Brain Rcs 493:205-2 11. of associative memory? Trends Neurosci 12:258-264. Scholfield CN (1978) A depolarizing inhibitory potential in neurones Hoffman WH, Haberly LB (1989) Do changes in synaptic excitation of the olfactory cortex. J Physiol (Lond) 275:547-557. or inhibition underlie generation of long-lasting bursting-induced late Sheldon PW, Aghajanian GK (1990) Serotonin (S-HT) induces IPSPs EPSPs in niriform cortex? Sot Neurosci Abstr 15:700. in pyramidal layer cells of rat pirifonn cortex: evidence for the in- Jung MW, Larson J, Lynch G (1990) Long-term potentiation of mono- volvement of a 5-HT,-activated . Brain Res 506:62-69. synaptic EPSPs in rat piriform cortex in vitro. Synapse 6~279-283. Traub RD, Wong RKS (I 983) Synchronized burst discharge in dis- Kanter ED, Haberly LB (1990) NMDA-dependent induction of long- inhibited hippocampal slice. II. Model ofcellular mechanism. J Neu- term potentiation in afferent and association fiber systems ofpiriform rophysiol 49:459-%7 1. cortex in vitro. Brain Res 525: 175-179. Tseng G-F, Haberly LB (1988) Characterization of synaptically me- Kanter ED, Haberly LB (199 I) Associative long-term potentiation in diated fast and slow inhibitory processes in piriform cortex in an in oiriform cortex slices requires GABA, blockade. Sot Neurosci Abstr vitro slice preparation. J Neurophysiol 59: 1352-l 376. i7:386. White G, Levy WB, Steward 0 (1990) Spatial overlap between pop- Kelso SR, Ganong AH, Brown TH (1986) Hebbian synapses in hip- ulations of synapses determines the extent of their associative inter- DocamDus. Proc Nat1 Acad Sci USA 83:5326-5330. action during the induction of long-term potentiation and depression. K&chum- KL, Haberly LB (1991) Fast oscillations and dispersive J Neurophysiol 64: 1186-l 198. propagation in olfactory cortex and other cortical areas: a functional Wigstrom H, Gustafsson B (1983) Facilitated induction of hippocam- hypothesis. In: Olfaction as a model system for computational neu- pal long-lasting potentiation during blockade of inhibition. Nature roscience (Davis J, Eichenbaum H, eds), pp 69-100. Cambridge, MA: 301:603-604. MIT Press. Wigstrijm H, Gustafsson B (1985) Facilitation of hippocampal long- Larson J, Wong D, Lynch G (1986) Patterned stimulation at the theta lasting potentiation by GABA antagonists. Acta Physiol Stand 125: frequency is optimal for the induction of hippocampal long-term 159-172. potentiation. Brain Res 368:347-350. Wigstrom H, Gustafsson B, Huang Y-Y, Abraham WC (1986) Hip- Levy WB, Steward 0 (I 979) Synapses as associative memory elements pocampal long-lasting potentiation is induced by pairing single affer- in the hippocampal formation. Brain Res 175:233-245. ent volleys with intracellularly injected depolarizing current pubes. Lupica CR, Dunwiddie TV (199 I) Differential effects of mu- and delta- Acta Physiol Stand 126:3 17-3 19. receptor selective opioid agonists on feedforward and feedback GA- Wilson MA, Bower JM (1988) A computer simulation of olfactory BAergic inhibition in hippocampal brain slices. Synapse 8:237-248. cortex with functional implications for storage and retrieval of olfac- Macrides F, Eichenbaum HB, Forbes WB (1982) Temporal relation- tory information. In: Neural information processing systems (An- ship between sniffing and the limbic 0 rhythm during odor discrim- derson DZ, ed), pp 114-l 26. New York: American Institute of Phys- ination reversal learning. J Neurosci 2: 1705-l 7 17. ics. Madison DV, Nicoll RA 7 1988) Enkephahn hyperpolarizes intemeu- rones in the rat hippocampus. J Physiol (Lond) 398: 123-l 30.